The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references and relies on source code from a third-party GitHub repository (https://github.com/siddhartha2303/cisco-sdwan-mcp) for its core logic, which is managed by a user other than the skill author.
[COMMAND_EXECUTION]: The skill executes a local Python script defined by the $SDWAN_MCP_SCRIPT environment variable to provide its tool functionality.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection because it ingests and processes data from external network devices (Cisco vManage).
Ingestion points: Data is ingested via tools such as get_alarms, get_events, and get_running_config (found in SKILL.md).
Boundary markers: No specific delimiters or instructions to ignore embedded content are defined for processing the vManage data.
Capability inventory: The skill possesses network communication capabilities and access to sensitive credentials (VMANAGE_PASSWORD).
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from the SD-WAN fabric before it is processed by the agent.

sdwan-ops